AI Content Recommendation Assistant Project

"""Recommend content by comparing transparent item descriptions."""

from __future__ import annotations

import pandas as pd
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity


def catalogue() -> pd.DataFrame:
    return pd.DataFrame([
        ("Python Foundations", "python programming functions files beginner"),
        ("Python Data Work", "python pandas data cleaning tables"),
        ("SQL Reporting", "sql queries joins aggregation reporting"),
        ("Power BI Dashboard", "power bi dax dashboard business metrics"),
        ("NLP Basics", "artificial intelligence natural language text classification"),
        ("RAG Systems", "artificial intelligence retrieval documents question answering"),
        ("Cloud Deployment", "cloud deployment containers monitoring"),
    ], columns=["title", "description"])


def recommend(items: pd.DataFrame, title: str, limit: int = 3) -> pd.DataFrame:
    matches = items.index[items["title"].str.casefold() == title.casefold()].tolist()
    if not matches:
        raise ValueError(f"Unknown title: {title}")
    matrix = TfidfVectorizer(ngram_range=(1, 2)).fit_transform(items["description"])
    scores = cosine_similarity(matrix[matches[0]], matrix)[0]
    order = [index for index in scores.argsort()[::-1] if index != matches[0]][:limit]
    result = items.loc[order, ["title"]].copy()
    result["similarity"] = scores[order]
    return result.reset_index(drop=True)


def main() -> None:
    print(recommend(catalogue(), "NLP Basics").round(3).to_string(index=False))
    print("Content similarity is not a claim about an individual learner's ability or preference.")


if __name__ == "__main__":
    main()